rpn - Reverse Polish Notation language

rpn

• is a math-oriented language inspired by Hewlett-Packard Reverse Polish Lisp (HP28S user manual is provided as a reference), it includes at least stack, store, branch, test, trig and logs commands
• is implemented as a a command-line calculator for GNU/Linux
• brings powerfull calculation facilities on floating point numbers with arbitrary precision, provided by GNU MPFR library
• uses that so cool reverse polish notation

Quick examples:

• easy calculation with stacked results

rpn> 1 exp 3 *
8.1548454853771357061
rpn> 2 sqrt
2> 8.1548454853771357061
1> 1.4142135623730950488
rpn> 

• programs and variables, eg same example as in HP28S Quick Reference:

rpn> << rot * swap 2 / chs dup sq rot - sqrt >> 'quadratic_solution' sto
rpn> 1 2 -3 quadratic_solution
2> -1
1> 2
rpn> vars
var 1: name 'quadratic_solution', type program, value << rot * swap 2 / chs dup sq rot - sqrt  >>

• local variables, always from the same reference:

rpn> << -> x y << x y + ln >> >> 'P' sto
rpn> 1 2 P
rpn> 1.0986122886681096914

• arbitrary precision (up to 0x7FFFFFFFFFFFFFFF bits with GNU MPFR !)

rpn> 256 prec 200 std
rpn> pi 3 * 4 / cos
-0.70710678118654752440084436210484903928483593768847403658833986899536623923105962592591087473479525356117497671223960240783675485777817360566073272153486395308799122357513534343724299243077135552002446
rpn> 

• Following objects are managed: floating numbers, symbols, strings, programs, plus language keywords (commands and flow controls)

4> 'symbol'
3> "string"
2> 12.3456
1> << -> n << 0 1 n for i i 2 * inv + next >> >>
rpn> 

• A GNU-readline-based interactive editor with autocompletion is provided.

keywords

general

keyword	description
nop	no operation
help	(or h or ?) this help message
quit	(or q or exit) quit software
version	show rpn version
uname	show rpn complete identification string
type	show type of stack first entry
default	set float representation and precision to default
prec	get float precision in bits when first stack is not a number, set float precision in bits when first stack entry is a number. ex: 256 prec
round	set float rounding mode. Authoerized values are: ["nearest", "toward zero", "toward +inf", "toward -inf", "away from zero"] round. ex: "nearest" round

real

keyword	description
+	addition
-	substraction
neg	negation
*	multiplication
/	division
inv	inverse
%	purcent
%	CH inverse purcent
^	(or pow) power
sqrt	square root
sq	(or sqr) square
mod	modulo
abs	absolute value

real representation

keyword	description
dec	decimal representation
hex	hexadecimal representation
bin	binary representation
std	standard floating numbers representation. ex: [25] std
fix	fixed point representation. ex: 6 fix
sci	scientific floating point representation. ex: 20 sci

test

keyword	description
>	binary operator >
>=	binary operator >=
<	binary operator <
<=	binary operator <=
!=	binary operator != (different)
==	binary operator == (equal)
and	boolean operator and
or	boolean operator or
xor	boolean operator xor
not	boolean operator not
same	boolean operator same (equal)

stack

keyword	description
swap	swap 2 first stack entries
drop	drop first stack entry
drop2	drop 2 first stack entries
dropn	drop n first stack entries
erase	drop all stack entries
rot	rotate 3 first stack entries
dup	duplicate first stack entry
dup2	duplicate 2 first stack entries
dupn	duplicate n first stack entries
pick	push a copy of the given stack level onto the stack
depth	give stack depth
roll	move a stack entry to the top of the stack
rolld	move the element on top of the stack to a higher stack position
over	push a copy of the element in stack level 2 onto the stack

string

keyword	description
->str	convert an object into a string
str->	convert a string into an object

branch

keyword	description
if	test-instructions
then	true-instructions
else	false-instructions
end	(end of if structure)
start	repeat instructions several times
for	repeat instructions several times with variable
next	ex: 1 10 start <instructions> next
step	ex: 1 100 start <instructions> 4 step

store

keyword	description
sto	store a variable. ex: 1 'name' sto
rcl	recall a variable. ex: 'name' rcl
purge	delete a variable. ex: 'name' purge
vars	list all variables
edit	edit a variable content

program

keyword	description
eval	evaluate (run) a program, or recall a variable. ex: 'my_prog' eval
->	load program local variables. ex: << -> n m << 0 n m for i i + next >> >>

trig

keyword	description
pi	pi constant
sin	sinus
asin	arg sinus
cos	cosinus
acos	arg cosinus
tan	tangent
atan	arg tangent
d	->r convert degrees to radians
r	->d convert radians to degrees

logs

keyword	description
e	exp(1) constant
log	logarithm base 10
alog	(or exp10) exponential base 10
log2	logarithm base 2
alog2	(or exp2) exponential base 2
ln	logarithm base e
exp	exponential
sinh	hyperbolic sine
asinh	inverse hyperbolic sine
cosh	hyperbolic cosine
acosh	inverse hyperbolic cosine
tanh	hyperbolic tangent
atanh	inverse hyperbolic tangent

default

Default float mode is 'std' with 20 digits

Default floating point precision is 128 bits

Default rounding mode is 'nearest'

Tests

Unit tests are given as txt files in the test subdirectory.

Use the command 'test' to run a test file, eg

# cd src_directory/
# rpn
rpn> "test/01-all.txt"
rpn> test
## ENTRY TEST
(...)
# real decimal PASSED
# real hex PASSED
# real binary PASSED
(...)

Test output is done on stdout and is not stacked in rpn.

generation and installation

At this time the software is proposed for GNU/Linux only and provided as a source code.

No binary package is provided, user must compile its version by following these 3 steps:

GNU GMP 6.1.2

• download and install GNU GMP v6.1.2 at https://gmplib.org/

GNU MPFR 3.1.5

• download and install GNU MPFR v3.1.5 from http://www.mpfr.org

rpn

• clone this project or download a zipped version from the release directory

  # git clone https://github.com/louisrubet/rpn.git
  

• configure and make

  # cd rpn/
  # ./configure && make
  

• install

  # sudo make install